Handy scripts to analyze the disk_usage output from Elasticsearch

disk_usage_by_field.sh

echo field, total, doc_values, inverted_index, points
curl -s $1 |
  jq -r '.
  | to_entries
  | .[]
  | select(.key != "_shards")
  | .value.fields
  | to_entries
  | .[]
  | "\(.key), \(.value.total_in_bytes), \(.value.doc_values_in_bytes), \(.value.inverted_index.total_in_bytes), \(.value.points_in_bytes)"'

disk_usage_otel_metrics_summary.sh

echo category, total, doc_values, inverted_index, points
curl -s $1 |
  jq -r '
  . | to_entries
  | .[]
  | select(.key != "_shards")
  | .value.fields
  | to_entries
  | map({
      field: .key,
      category: (if .key | startswith("_") then "meta"
        elif .key == "@timestamp" then "@timestamp"
        elif .key | startswith("attributes") then "attributes"
        elif .key | startswith("resource.attributes") then "attributes"
        elif .key | startswith("metrics") then "metrics"
        else "other"
        end
      ),
      total_in_bytes: .value.total_in_bytes | tonumber,
      inverted_index_total_in_bytes: .value.inverted_index.total_in_bytes | tonumber,
      doc_values_in_bytes: .value.doc_values_in_bytes | tonumber,
      points_in_bytes: .value.points_in_bytes | tonumber
    })
  | group_by(.category)
  | map({
      category: .[0].category,
      total_in_bytes: map(.total_in_bytes) | add,
      inverted_index_total_in_bytes: map(.inverted_index_total_in_bytes) | add,
      doc_values_in_bytes: map(.doc_values_in_bytes) | add,
      points_in_bytes: map(.points_in_bytes) | add
    })
  | .[]
  | "\(.category), \(.total_in_bytes), \(.doc_values_in_bytes), \(.inverted_index_total_in_bytes), \(.points_in_bytes)"'